Circumventing rain‐related errors in scatterometer wind observations

Satellite scatterometer observations of surface winds over the global oceans are critical for climate research and applications like weather forecasting. However, rain‐related errors remain an important limitation, largely precluding satellite study of winds in rainy areas. Here we utilize a novel technique to compute divergence and curl from satellite observations of surface winds and surface wind stress in rainy areas. This technique circumvents rain‐related errors by computing line integrals around rainy patches, using valid wind vector observations that border the rainy patches. The area‐averaged divergence and wind stress curl inside each rainy patch are recovered via the divergence and curl theorems. We process the 10 year Quick Scatterometer (QuikSCAT) data set and show that the line‐integral method brings the QuikSCAT winds into better agreement with an atmospheric reanalysis, largely removing both the “divergence bias” and “anticyclonic curl bias” in rainy areas noted in previous studies. The corrected QuikSCAT wind stress curl reduces the North Pacific midlatitude Sverdrup transport by 20–30%. We test several methods of computing divergence and curl on winds from an atmospheric model simulation and show that the line‐integral method has the smallest errors. We anticipate that scatterometer winds processed with the line‐integral method will improve ocean model simulations and help illuminate the coupling between atmospheric convection and circulation.